Closed Bug 744689 Opened 12 years ago Closed 12 years ago

QA and deploy BrowserID train-2012.04.11 to production

Categories

(Cloud Services :: Operations: Deployment Requests - DEPRECATED, task)

x86_64
Linux
task
Not set
normal

Tracking

(Not tracked)

VERIFIED FIXED

People

(Reporter: lloyd, Assigned: petef)

References

Details

(Whiteboard: [qa+])

Version: 20b86fd (0.2012.04.11.1)

Tests pass: http://travis-ci.org/#!/mozilla/browserid/builds/1070738

ChangeLog including issues resolved: 
https://github.com/mozilla/browserid/blob/train-2012.04.11/ChangeLog#L1-15

[QA] Suggested additional areas of focus for QA:
  * verify no regression on security issue 741814
  * all issues in changelog
  * extensive regression testing of all RPs (show/booze/beer).
  * coordinate with lloyd, and subsequent to that testing of the new API via 123done.org

[ops] deployment issues:
  * let's update nginx config to gzip resources on the way out, when appropriate - something like:

    gzip  on;
    gzip_disable "MSIE [1-6]\.(?!.*SV1)";
    gzip_proxied any;
    gzip_types text/html application/javascript text/css;
please deploy 0.2012.04.11.3 to stage.  This includes the following hotfixes:

  * (hotfix 2012.04.12) fix broken string, "is this your computer" was broken into two fragments: #1425
  * (hotfix 2012.04.12) return 400 rather than 500 for invalid params to stage_user or stage_email: #1429
  * (hotfix 2012.04.12) fix /signup page on browserid.org: #1429

#1429 also went into production as a hotfix to train-2012.03.28

ver 0.2012.04.11.3 - sha 51c9909 - https://github.com/mozilla/browserid/tree/train-2012.04.11
deployed to stage, along with nginx gzip config.
Verified the following ver.txt in Stage:
51c9909 0.2012.04.11.3 - fix broken string that would have caused 'is this your computer' to not make sense for non-en users
locale svn r104263

QA accepts this train for testing.
curl -D - --compressed https://... should show 'Content-Encoding: gzip' somewhere in the response headers from the server.
backend and frontend tests (across mac/win7/win8/ubuntu/android2.2/android4.0)
complete
  - noted some flakiness in win32/safari for frontend tests:
    https://github.com/mozilla/browserid/issues/1441
- have one defect that probably requires a hotfix:
    https://github.com/mozilla/browserid/issues/1440
- I have another case where I get 400 Bad request: requires authentication
  for /wsapi/cert_key. Am looking for concrete STR before filing issue.
- load_gen tool needs some updating in order to work with current train:
  https://github.com/mozilla/browserid/issues/1436
(In reply to John Morrison [:jrgm] from comment #5)
> - I have another case where I get 400 Bad request: requires authentication
>   for /wsapi/cert_key. Am looking for concrete STR before filing issue.
  Now https://github.com/mozilla/browserid/issues/1445

> - load_gen tool needs some updating in order to work with current train:
>   https://github.com/mozilla/browserid/issues/1436

Done and verified. 

BFT and bug verify continue.
petef, please deploy version 0.2012.04.11.7 - sha 1613144c4e

this includes the following fixes: https://github.com/mozilla/browserid/blob/1613144/ChangeLog#L18-22
(In reply to Lloyd Hilaiel [:lloyd] from comment #7)
> petef, please deploy version 0.2012.04.11.7 - sha 1613144c4e
> 
> this includes the following fixes:
> https://github.com/mozilla/browserid/blob/1613144/ChangeLog#L18-22

browserid-server-0.2012.04.11-7.el6_104263.x86_64.rpm deployed to stage & clientX load-gen boxes
Verified the following push to Stage:
1613144 0.2012.04.18.7 - fix issue #1449 - whereby user logs in and immediately sees an authentication error
locale svn r104263

For the record, noting Hot Fixes here (as of 4/18/2012):
Hot Fix 1:
1429: (hotfix 2012.04.12) return 400 rather than 500 for invalid params to stage_user or stage_email
1430: (hotfix 2012.04.12) return 400 rather than 500 for invalid params to stage_user or stage_email
1425: (hotfix 2012.04.12) fix broken string, "is this your computer" was broken into two fragments
Hot Fix 2:
1442: (hotfix 2012.04.16) fix API regression that would cause javascript error when .get() invoked without second arg
1436: (hotfix 2012.04.16) update load_gen to new server apis that require an `ephemeral` argument
Hot Fix 3:
1440: (hotfix 2012.04.17) fix broken reset password flow - button was non-responsive in dialog
1445: (hotfix 2012.04.17) mitigate errors seen when adding a secondary email to an acct with only primary emails
Hot Fix 4:
1449: (hotfix 2012.04.18) fix error where under certain conditions user could see an error immediately after authenticating
Status: NEW → ASSIGNED
Whiteboard: [qa+]
Scratch the verification portion spelled out in Comment 9. OPs is rebuilding now...
Pushed again and verified:
1613144 0.2012.04.18.7 - fix issue #1449 - whereby user logs in and immediately sees an authentication error
locale svn r104263

QA moves on with testing this current push to Stage...
Complete:
- confirmed that js/css static asserts are served with gzip encoding (when
  accept-encoding: gzip,deflate)
- all issues in changelog verifications complete
- backend and fronted tests pass
- regression testing of (show/booze/beer) complete on
  windows(ie8/ie9/safari/opera/chrome/firefox),
  osx(safari/opera/chrome/firefox), ubuntu(chrome/firefox/opera),
  android2.2[droid2](stock/ff11/aurora),
  android4[asus-tablet,google-nexus](stock/ff11/aurora).

To complete:
- preliminary check that locales have content complete. Will need to check the
  final l10n bits tomorrow.
- need to find a time to run a longer load test
- working with 123done
- pending hotfix
HotFix 5 just pushed to Stage:
5fc0c75 bump rpm release number
locale svn r104128

QA to perform the following over the next two days:
Sanity testing
iOS testing
HotFix verification
long-term load test
HotFix 5:
1464: [ie8/ie9] IE reporting a problem with test check_registration: user validation with mustAuth result
1485: [IE8] error dialog shows up with solid black background (i.e., not just dimmed)
Various sec stuff
Load Test is in progress:
client5:
bin/load_gen -s https://stage-browserid.services.mozilla.com -o -m 250000 -u 1/250
Noting two Hot Fixes for the next train:
Hot Fix 1487: Hotfix 1484 chrome exception signup
1484: [Chrome] exceptions thrown on /signup page when cookies disabled

Hot Fix 1494: Hotfix 1390 min browser reqs
1390: IE6 and IE7 errors in communication frame

And one possible Hot Fix for the next train:
1418: signup/signin/forgot pages from main site should check for cookies
(related to 1484)
- Load test ran fine for ~30 hours. Two errors during the run, both look like they 
came from the load balancer. 
- all testing complete

QA signs off on train-2012.04.11.
Although, still need to get the final up to date L10n bits. Petef, do you have time to do this?
petef, final locales are in.  please deploy 2012.04.11.9 - sha 57ea50a37 - https://github.com/mozilla/browserid_private/tree/train-2012.04.11-with-sec-hotfix
diresworb.org/ver.txt is 
57ea50a bump version number with inclusion of final shipping locales.
locale svn r104778
- confirmed basic functional use of the new bits on stage
- no problems loading all locales dialog.js in FF
- welsh is apparently not picked up in this build (GH-1503)
Looks good to me
2012.04.11.9 has been pushed to Production (PHX1 and SCL2):
https://intranet.mozilla.org/Services/Ops/ChangeWindow_20120426

Details:
57ea50a bump version number with inclusion of final shipping locales.
locale svn r104778
Status: ASSIGNED → RESOLVED
Closed: 12 years ago
Resolution: --- → FIXED
QA verified the push to Production.
Some non-blocking issues came up that will be filed separated and are not considered blockers.
Status: RESOLVED → VERIFIED
Blocks: 749756
Blocks: 754877
No longer blocks: 754877
You need to log in before you can comment on or make changes to this bug.